The release remedy reads a version out of an issue's body, so an ordinary issue is told a published release finishes it #2784

What is observed

scripts/check-issue-links.py now reads a release version out of an issue's text and, when that issue's declaring PR has merged with no open one left, sends the reader to that artifact instead of to the merge:

what finishes this issue is the published vX.Y.Z, not that merge … uv run --no-sync python3 scripts/check-release-published.py vX.Y.Z is the reading: green across the platform matrix is what says the work is done, and this issue may be closed on that reading

For a release issue that is right. The version is read from both the title and the body, and a body is where an issue says what it was seen in, not what it is.

Measured over the repository's last 200 issues: 34 name a version, 26 of them in the title and 8 only in the body. Driven through judge_issues with three of those real texts, each gets the artifact sentence while its declaring PR merged work about something else entirely:

Why it is worse than the sentence it replaced

The version each of those bodies names is already published. check-release-published.py v0.3.41 therefore exits 0, and the row tells the reader the issue "may be closed on that reading" — a green reading about a release the issue was never about, certifying completion of work that is not done. The older sentence named no reading at all and left the reader to find the right one; this one hands them a reading that answers yes for the wrong reason.

What is expected

The artifact remedy belongs only where the issue's own text declares the version it is about. The title is that place: a title says what an issue is, a body says what it was seen in.

What the fix must satisfy

• An issue naming a version only in its body keeps the remedy it had before — "either close this issue with the reading that says the work is done, or state here what it still leaves" — and the artifact command does not appear in its row.
• An issue naming the version in its title still gets the artifact sentence, unchanged.
• Both are pinned through judge_issues, the function the report actually calls, not only through the helper under it.
• The residual limit is stated rather than implied: an ordinary issue whose title names a version would still take the artifact sentence.